Coral bleaching and climate change; an investigation on thermal stress on coral reefs.

Ruben J van Hooidonk, Purdue University

Abstract

This dissertation addresses important issues in predicting coral bleaching. The goal of this study is to produce forecasts of bleaching that have a good and quantifiable quality. Thermal coral bleaching is a major, global threat to coral reefs. Coral bleaching is the potentially lethal condition where heterotrophic corals become white due to a heat stress induced decrease in concentration of their zooxanthellae – photosynthetic dinoflagellate symbionts (Symbiodinium spp.). Techniques have been developed utilizing observational estimates of sea surface temperature to make quantitative predictions of coral reef bleaching induced by thermal stress. This basic approach is widely used and forms the foundation of predictions for the global demise of coral reef ecosystems within the next 20–40 years, as a byproduct of anthropogenic climate change. Furthermore, predictions of thermally induced coral bleaching such as predictions based on Degree Heating Weeks (DHW) are cited as an important tool in managing reefs. Through interactions with other coral reef scientists and a literature study it became apparent that there are five problems which, if solved, would lead to improvements in forecasts about coral reef health. One problem is that (1) forecasts are not verified in a quantitative matter. Other issues in existing predictive techniques are (2) the use of one global threshold for all locations with reefs, (3) coarse temporal resolution of future sea surface temperature data used, (4) uncorrected errors in high frequency variability in sea surface temperature data from general circulation models used, and (5) the small number of general circulation models used in studies predicting future bleaching. In this dissertation I address all of these issues. I apply a forecast verification technique commonly used in meteorology to predictions of coral bleaching to verify and quantify the quality of the forecasts. With this verification technique it is possible to (a) measure improvements in the predictive technique, (b) quantitatively compare techniques, and (c) identify locality specific thresholds. I show that predictive techniques can be improved by tuning thresholds of the predictor per reef location instead of using one global threshold. This has the added advantage that for each reef the optimal thermal stress threshold becomes known allowing us to identify reefs at risk and more resilient reefs. This approach adds value to existing predictive techniques and enables comparison with other techniques. In this work I use weekly data of thermal stress experienced by corals where previous studies have used monthly data, this increases the quality of the forecasts. I also improve the quality of the forecasts by implementing a bias correction to the sea surface temperatures from general circulation models. This correction reduces difference in short term variability between observations and modeled sea surface temperatures. Projections of future thermal stress are also improved by using a large ensemble of models compared to two to four models used in previous studies. The ensemble approach is known to produce better predictions in other climate parameters.

Degree

Ph.D.

Advisors

Huber, Purdue University.

Subject Area

Ecology|Climate Change|Atmospheric sciences|Remote sensing

COinS